[K12OSN] one year performance graphs and statistics for two server setups we use
Les Mikesell
lesmikesell at gmail.com
Sat Jun 21 19:59:16 UTC 2008
Nadav Kavalerchik wrote:
> Cacti already monitor SNMP. i was looking for the exact SNMP strings you
> might think help me monitor router's bottle-necks , NIC's IO buttle-necks,
> cpu wait for IO, squid wait for...
> (actually, i monitor some hardware with SNMP. but i did not include it in
> the report.)
>
Cacti is easier to set up, but opennms is more comprehensive and does
more by default. It will have default graphs for CPU use that
color-code the user/system/wait times. I don't think there is anything
specific for NIC buffers but it will graph interface errors, total
current connections, and connections being opened in in/out bound
directions as well as interface bandwidth. And it can notify you via
email, sms, a jabber chat room, etc. when something goes wrong.
But there's probably only a few things that matter - one will be total
outbound bandwidth for screen-intensive things, another will be CPU use
for things like flash that all run on the server CPU. These are easy to
see even with cacti. Another is disk head contention where multiple
users/processes all are waiting for something in different locations on
the same disk. That one is harder to diagnose - and likely the most
common bottleneck.
--
Les Mikesell
lesmikesell at gmail.com
More information about the K12OSN
mailing list